DNA methylation and histone modiﬁcations are epigenetic regulatory

In 1939, Conrad Hal Waddington published his book ”An Introduction to Modern Gen-etics” in which he ﬁrst introduced and deﬁned the phrase ”EpigenGen-etics”.^64,65 Thus, epi-genetics is almost 15 years older than Watson’s discovery of the DNA-doublehelix.⁶⁶ Waddigton created the phrase ”Epigenetics” as a name for a model to describe how genes might interact with their surroundings to produce a phenotype. Over the decades, the deﬁnition of ”Epigenetics” developed and changed. In one example, epigenetics was described by Russo et al. as ”The study of mitotically and/or meiotically heritable changes in gene function that cannot be explained by changes in DNA-sequences.”⁶⁷ An-other way of describing epigenetics was made by Adrian Bird. He deﬁned epigenetics as

”The structural adaption of chromosomal regions so as to register, signal or perpetuate altered activity states”.⁶⁸ The best understood mechanisms responsible for structural adaption are DNA methylation and histone modiﬁcations, both altering gene expression during development and cancer progression.⁶⁹ Of course, other epigenetic mechanisms are known which also inﬂuence gene-expression such as miRNA mediated gene silencing.

However, due to their importance in this work, the explanations in the next sections focus on DNA methylation and histone modiﬁcations.

1.4.1 DNA methylation and histone modiﬁcations

DNA methylation was ﬁrst discovered by Rollin Hotchkiss in calf thymus DNA in 1948.⁷⁰ Typically, it occurs at CG dinucleotides (CpGs) in the human genome. The enzymes re-sponsible for DNA methylation are DNA methyltransferases (DNMTs), ﬁrst discovered in 1975.⁷¹ DNMT1,⁷² responsible for maintenance of methylation during cell cycle, and DNMT3a and 3b^73,,⁷⁴ responsible for de novo methylation, are the most important DN-MTs in mammals. Genomic regions which can be methylated are e.g. located within the genes’ promoter regions. There, CpG-rich regions, so called CpG islands, are regulatory sites capable of inhibiting gene transcription when methylated.⁷⁵ CpG-islands are char-acterized by a GC-content greater than 50 % with an observed to expected CpG ratio of greater than 60 % in a minimum of 200 bp region.⁷⁶ Other methylation-dependent events are genomic imprinting⁷⁷ or X-chromosome inactivation⁷⁸ causing chromosomal closure.

However, the majority of methylation within the human genome was not discovered in CpG-islands, but in CpG-positions within CpG-poor regions (1 CpG per 100 bp).^79,80In a genome wide approach, Brenet et al. recently discovered that ﬁrst exon methylation is linked to transcriptional silencing.⁷⁹ Unfortunately, they were not able to present a model for the mechanism of intragene methylation-mediated gene expression, yet.

An additional means of DNA modiﬁcation was discovered recently, called CpG hydroxymethylation. It is caused by oxidation of methylated CpGs by TET1.⁸¹ Although little is known about CpG hydroxymethylation, it might contribute to gene regulation and will inﬂuence epigenetic research in the future.

Other important epigenetic features are histone modiﬁcations. Due to their ca-pability to determine if DNA is accessible for transcription or not, they have a major impact on gene expression, too. First described by Albrecht Kossel in 1884,⁸² histones are a protein class, containing histone H1 and the core histones H2A, H2B, H3, and H4. A nucleosome is formed by an octamer of the core histones and 146 basepairs of DNA.⁸³ Together with H1, the nucleosomes form the chromatin structure where DNA is condensed up to 10000-fold.⁸⁴ However, to allow processes like transcription or repli-cation, this structure needs to change dynamically from a condensed state to locally open states. These dynamic changes are aﬀected by three covalent histone modiﬁca-tions on the aminoterminal ends of the core histones: acetylation, phosphorylation and methylation. Besides some exceptions in yeast and drosophila,^85,86 histone acetylation is mainly linked to transriptional activation.⁸⁷ One important acetylation site is Lysine 14 at the aminoterminal end of histone 3 (H3K14).⁸⁸ By introducing or removing acetyl-groups, histone acetyltransferases (HATs)⁸⁹ and histone deacetylases (HDACs)⁹⁰ change the aﬃnity to DNA and adjacent nucleosomes.

Histone phosphorylation is also associated with transcriptional activation.⁹¹ How-ever, the mechanism is not well understood, yet. Similar to acetylation, phosphorylation might reduce the aﬃnity to DNA and nucleosomes by introducing a negative charge. It

was also shown that phosphorylation stimulates HAT activity.⁹²

Histone methylation can be associated either with transcriptional silencing (H3K9 di- or trimethylation)⁹³ or activation (H3K9 monomethylation⁹³ or H3K4 trimethyla-tion⁹⁴). One eﬀect of methylated histones is, for instance, the binding of heterochromatin speciﬁc proteins like HP1 in silencing mechanisms.⁹⁵

1.4.2 Epigenetic regulatory mechanisms

DNA methylation-dependent gene regulation is mainly mediated via two mechanisms.

The ﬁrst mechanism involves DNA methylation-dependent transcription factors includ-ing the most prominent ones, AP-2⁹⁶ and Sp1.⁹⁷ Both do not bind when the binding sites are methylated, leading to a reduction of gene expression.

The second mechanism comprises the connection between DNA methylation and chromatin structure, established by proteins binding to methylated DNA. One of the proteins is MeCP2⁹⁸ which has a methylation binding domain (MBD-domain)⁹⁹ and a transcriptional repression domain (TRD-domain). Thus, the protein is able to re-cruit a corepressor complex, consisting of mSin3A and HDACs, to methylated DNA.

By deacetylation the chromatin changes to the condensed state leading to transcript-ional silencing.¹⁰⁰ Besides of MeCP2, other MBD-containing proteins like MBD2 are able to recruit HDACs to methylated DNA.¹⁰¹ The functional interaction between DNA methylation and histone modiﬁcations were also veriﬁed by studies showing that HDAC inhibitors like Trichostatin A are able to relief MBD-containing protein mediated trans-criptional silencing.102,101,103 Opposing this mechanism, chromatin structure can also inﬂuence DNA methylation. By trimethylating H3K27, polycomb group proteins, such as EZH2, induce transcriptional silent chromatin.¹⁰⁴ Furthermore, they are able to re-cruit DNMTs to the transcriptional silent sites¹⁰⁵ leading to methylated DNA.

In contrast to the MBD-containing proteins, the zinc ﬁnger containing proteins ZBTB33 (Kaiso)¹⁰⁶ and ZBTB4 are two members of another group of transcription fac-tors. Their zinc ﬁnger motifs show higher aﬃnity to methylated DNA sequences than to unmethylated DNA in vitro and thereby might repress transcription.¹⁰⁷ Addition-ally, ZBTB33 might be able to suppress gene expression also methylation-independently through extra zinc ﬁnger binding motifs, called BTB/POZ-domains.¹⁰⁶ The described functions of ZBTB33 include, for instance, the recruitment of the N-CoR repressor com-plex, a protein complex which promotes histone deacetylation leading to silent chro-matin.¹⁰⁸ However, in a recent publication it was shown that the methylation-dependent binding might only play a minor role in the function of ZBTB33.¹⁰⁹ By analyzing ENCODE-data¹¹⁰ the authors rather identiﬁed an association of ZBTB33-binding and actively expressed genes.

CTCF is also a zinc ﬁnger containing DNA-binding factor, which can bind to CpG-containing sites and to sites without CpGs, too.¹¹¹ Contrary to ZBTB33 and ZBTB4, methylation of the CpG-containing binding sites reduces binding aﬃnity.¹¹² CTCF can

either function as a transcriptional repressor¹¹³ or as a transcriptional activator.¹¹⁴ The most important function of CTCF is to be an insulator protein being necessary to block promoters from the inﬂuence of functionally-independent enhancers.¹¹⁵ CTCFL (BORIS), a paraloque of CTCF, is able to bind the same DNA motifs as CTCF.¹¹⁶ De-tected in several cancer cells,¹¹⁷ CTCFL is thought to interfere with CTCF-binding and to function as an antagonist to CTCF.¹¹⁸

1.4.3 The impact of epigenetic mechanisms in cancer development

Epigenetic mechanisms play a major role in the formation and maintenance of all types of cancers. In 1983, it was detected that tumor tissues have a globally reduced methylat-ion content in comparison to corresponding normal tissues.^119,120 The results were later veriﬁed by high throughput DNA methylation analysis techniques like microarrays.¹²¹ This feature is called hypomethylation and is mainly found in gene poor areas¹²² but sometimes, it can also occur at CpG-islands in promoters¹²³ when growth-related genes are aﬀected. The PAX2 gene promoter e.g. was found to be hypomethylated in en-dometrial cancers, but not in normal enen-dometrial tissues.¹²⁴ Besides growth activation, hypomethylation also plays a role in chromosomal instability^125,126 and chromosomal rearrangement,¹²⁷ which is for example observable in the reactivation of transposons.¹²⁸ A further consequence of this, is loss of imprinting which can be seen, for instance, for the insulin-like growth factor 2 in colorectal cancers (IGF2).^129,130 During cancer deve-lopment DNA hypomethylation increases.¹³¹ In contrast to hypomethylation, hyperme-thylation mainly occurs in gene promoters. In diﬀerent cancer types, the promoters of the genes BRCA1,¹³² p16,^133,134 E-cadherine¹³⁵ and VHL¹³⁶ can be methylated, leading to their down-regulation. The hypermethylation pattern is speciﬁc for the cancer type and increases during cancer development.^137,138

For colorectal cancer, a particular hypermethylation pattern is described as CpG-island methylator phenotype (CIMP). The CIMP-phenotype, ﬁrst described by Toyota et al. in 1999, is characterized by hypermethylation of several promoter CpG-islands associated with inactivation of tumor suppressor genes.¹³⁹ Toyotaet al. suggested a list of 30 diﬀerent CpG-islands to distinguish tumors in CIMP positive or CIMP negative tumors. Since its discovery, the CIMP-phenotype was associated with epidemiological features, like age, gender or location of the tumor, as well as genetic features, like MSI, KRAS and BRAF mutations.^140,141 For example, when the hMLH1-promoter is among the methylated loci, CIMP positive tumors mainly have an MSI phenotype. But, in contrast to MSI or other well known phenotypic subtypes of cancers, CIMP still is under controversial discussion and not accepted by the entire scientiﬁc society.^142,143 The main reason is that the sites determining CIMP phenotype were not standardized, yet. For the 30 diﬀerent CpG-islands of the analyzed CIMP positive tumors a variety of diﬀerent methylation patterns is observable.¹³⁹ As a consequence, each individual tumor might have a diﬀerent expression pattern of the 30 genes whose promoters are tested. Also, the

mechanism for CIMP tumor development is still unknown. Therefore, there is no clear argument supporting the relevance of the tested CpG-islands for predicting the CIMP status.

A further example of promoter methylation leading to silenced gene expression is the retinoblastoma gene (RB).^144,145Silenced RB cannot contribute to a protein complex containing the chromatin remodeling proteins SWI/SNF,¹⁴⁶ HDACs,¹⁴⁷ polycomb class epigenetic silencing proteins¹⁴⁸ and DNMT1.¹⁴⁹ The complex is necessary to silence RB targets as e.g. the cell cycle activator geneCyclin E¹⁵⁰by changing the chromatin struc-ture. This is an important example illustrating that DNA methylation is tightly linked to chromatin structure in cancer cells, too. The ﬁndings are further supported by changes in histone patterns like global loss of H4K16 acetylation or H4K20 trimethylation. These modiﬁcations were identiﬁed as common characteristics of cancer cells.¹⁵¹ Additionally, the expression of histone modifying enzymes diﬀers between healthy tissues and cancer tissues and also between the cancer types.¹⁵² As these examples illustrate, epigenetic fea-tures play an important role in cancer research. Since these feafea-tures are also connected to cancer development or the response of cancer tissues to medication, epigenetics also raises importance in clinical setting as prognostic, diagnostic and predictive markers.

1.5 Cancer treatment

1.5.1 Standard treatment options for cancer patients

Cancer can be treated by various therapeutic approaches. The most common approach is a surgery where the tumor and close lymph nodes are removed.¹⁵³ However, if the tumor is located in a non-accessible region or if there is a high estimated risk that the tumor relapses, further treatment options are applied. Radiation therapy is a means of killing tumor cells by electromagnetic or particle radiation.¹⁵⁴ The advantage of this is that ra-diation can be applied directly to the tumor with optimized intensities for the individual patient. The disadvantage is that surrounding healthy tissues might be harmed by the radiation, too, leading to a variety of side eﬀects like hair loss, damaged organs or recur-ring cancer. A further therapy approach is chemotherapy. Chemotheapy is used to kill fast proliferating cells, like tumor cells, by compounds inﬂuencing cell proliferation.¹⁵⁵ One example is the FOLFIRI therapy in colorectal cancers which combines folinic acid, 5-ﬂuorouracil and irinotecan hydrochloride.¹⁵⁶ Folinic acid interferes with nucleotide synthesis processes. 5-ﬂuorouracil is a substitute of Cytosine and Thymine during DNA polymeration processes. Irinotecan hydrochloride is a topoisomerase inhibitor. Due to their impact on DNA replication and cell cycle processes, these compounds do not only kill tumor cells but also other fast proliferating cells like blood cells. This leads to side eﬀects which are for example fatigue, digest problems or hair loss. To circumvent these problems, new therapeutics were tested which inhibit cell proliferation by targeting a tumor speciﬁc feature. These targeted therapeutics might be able to only aﬀect the

tumor and not to inﬂuence normal tissue cells.

1.5.2 EGFR-targeted therapies

Over-expression of the EGF-receptor is a feature, widely spread among diﬀerent tumor types. EGFR-inhibition will immediately lead to inhibition of cell growth and diﬀeren-tiation. Therefore, several compounds targeting EGFR were developed.

Cetuximab Also called Erbitux^r, the most important therapeutic to target EGFR is Cetuximab. Cetuximab is a chimeric human/mouse monoclonal antibody. The mouse precursor antibody mAB225 was developed by Gill et al. in 1984.¹⁵⁷ They already discovered that mAB225 is able to inhibit EGF-binding and to block EGFR autophos-phorylation in A431 cells, an epidermoid carcinoma cell line with EGFR over-expression.

In subsequent experiments, it was investigated that mAB225 also inhibits breast can-cer,^158,159 colon cancer,^160,161 renal cancer¹⁶²and prostatic cancer cell lines.^163,164 Exper-iments in xenografts^165,166 showed similar results and paved the way for the ﬁrst phase I clinical trial. Patients with squamous cell carcinomas of the lung were treated with

111In-labeled mAB225 to test for toxicity and to visualize the tumor.^167,168 The great advantages of the antibody therapy was that no toxicity was observed when patients were treated with high doses and that the antibody targeted the lung tumor cells di-rectly. This was observed by radiolabel-dependent visualization three to ﬁve days after injection.

In 1993, the mAB225 was chimerized by Naramura et al.¹⁶⁹ to avoid human anti-mouse antibody response (HAMA).¹⁷⁰ Similar to mAB225, the mouse-human chimeric C225, now called Cetuximab, also inhibited cell growth in several cancer cell lines and xenografts.171,172,173,174As a consequence, numerous phase II and phase III clinical trials were performed. In most of the cases Cetuximab was combined with other drugs or treatment options like radio- or chemotherapy. In a phase III clinical trial from 2005, it was seen that initial treatment of metastatic colorectal cancer with Cetuximab plus chemotherapy (FOLFIRI) reduced the risk of disease progression by 15 % in compari-son to treatment with FOLFIRI alone (CRYSTAL-study).¹⁷⁵ In this study, it was also seen that KRAS-wildtype patients responded signiﬁcantly better to chemotherapy plus Cetuximab than KRAS-mutant patients. However, also among the KRAS-mutant pa-tients a subset of papa-tients responded better to Cetuximab plus FOLFIRI (odds ratio (95

% CI) = 0.80 (0.44 - 1.45), odds ratio > 1: beneﬁt from Cetuximab plus FOLFIRI).

Finally, as a result of this study, Cetuximab was approved in the USA and in Europe for the treatment of metastatic colorectal cancers e.g. in combination with FOLFIRI when the tumor highly expresses EGFR. Yet, this treatment option is only valid for KRAS-wildtype patients.

Besides Cetuximab, other antibodies were developed to interfere with EGFR-function in colorectal cancer. One example is Panitumumab (Vectibix^r) which was

also approved for monotherapy and in combination with chemotherapeutics. However, small molecules targeting the intracellular tyrosine kinase domain of the EGFR are also promising. The most widely used ones are Erlotinib and Geﬁtinib.

Erlotinib and Geﬁtinib Erlotinib was shown to induce cell cycle arrest and apoptosis in cancer cell lines.¹⁷⁶ Also, Geﬁtinib was shown to inhibit the MAPK-pathway in high EGFRexpressing cancer cells.¹⁷⁷ Both compounds were subjected to several clinical tri-als. In contrast to Cetuximab, Erlotinib as well as Geﬁtinib failed to increase response in combination therapies with chemotherapeutics.178,179,180,181 However, as a monotherapy, Erlotinib succeeded to increase the life-span of NSCLC-patients.¹⁸² Geﬁtinib also led to a better outcome, but only in patients harboring EGFR mutations.¹⁸³ The SATURN-trail conﬁrmed an increased eﬀect for patients with EGFR mutations, too.¹⁸⁴ As a consequence of these studies, FDA and EMA approved Erlotinib for the treatment of non-small cell lung cancer and pancreatic cancer. Geﬁtinib was also approved for the treatment of non-small-cell lung cancer, if the tumor comprises an activating EGFR mutation. However, none of the compounds were approved for treatment of colorectal cancer until now. But, phase II clinical trials are in progress.185,186,187,188

Targeted therapies gain importance in clinical treatment procedures. Patients are stratiﬁed with biomarkers likeKRAS- orEGFRmutations to increase the response rates of such therapies. However, other biomarkers will be necessary to further optimize the stratiﬁcation procedure for example to exclude patients who will not respond, although they are KRAS-wildtype.

1.5.3 AREG and EREG expression are predictive markers for EGFR-targeted therapies

Due to their inﬂuence on EGFR-activation leading to activated PI3K- and MAPK-pathways, AREG and EREG might interfere with inhibitory mechanisms of EGFR in-hibitors. Because of this, they were considered as potential predictive markers to evaluate the outcome of EGFR-targeted therapies. Several studies were e.g. performed in non-small cell lung cancers. On the one hand, Ishikawaet al. and Masagoet al. showed that AREG expression correlates with a poor response to Geﬁtinib in NSCLC-patients.^189,190 On the other hand, Yonesakaet al. found out that Geﬁtinib as well as Cetuximab led to a higher growth inhibition in high AREG expressing NSCLC-cell lines compared to low AREG expressing NSCLC-cell lines.¹⁹¹ They veriﬁed the cell line data by immunohis-tochemistry analyses in patient-derived NSCLC-tumors. High AREG-staining indicated a stable disease after Erlotinib or Geﬁtinib treatment, whereas low AREG expressing patients mainly showed progressive disease. Vollebergh et al. also veriﬁed these results by testing the sera of a large patient cohort. High AREG levels were associated with a signiﬁcantly decreased risk of death after Erlotinib or Geﬁtinib treatment.¹⁹² Interest-ingly, in the same study, low TGF-α levels correlated with a better prognosis than high

TGF-α levels.

In metastatic colorectal cancers, AREG and additionally EREG were also identi-ﬁed as predictive markers for the outcome of EGFR-targeted therapies. In a 110-patient-study, Khambata-Ford et al. observed that highAREG mRNA and highEREG mRNA expression levels were mainly detected in patient biopsies which responded to Cetuximab therapy. They also showed that high AREG and EREG mRNA expression correlates with a longer progression-free survival time. Finally, they veriﬁed previous data illustrat-ing that KRAS-wildtype patients responded better to Cetuximab than KRAS-mutant patients.¹⁹³ However, in their study, the mutation status of the patients did not inﬂuence the potential of AREG and EREG as predictive markers. In contrast to these results, Ja-cobset al. observed that theKRAS mutation status is indeed a criterion for the potential

Im Dokument The epigenetic regulation of the EGF-receptor ligands Amphiregulin and Epiregulin and its impact on the outcome of EGFR-targeted therapies (Seite 19-28)